An automated system that identifies work at height and the fastening state of safety hooks using wearable sensors was developed to prevent falls from height (FFH). This system estimates the altitudes of workers based on the atmospheric pressure measured by a barometer and acceleration and gyroscopic signals from an inertial measurement unit (IMU). The fastening state of the safety hooks of workers at height is determined with the data collected by the IMU sensor and machine learning algorithms. Although researchers have tried to detect unsafe work conditions and unsafe behaviors at height, the complicated tasks and dynamic work conditions have discouraged them from establishing precise methodologies for effective and timely detection. To validate the system of this study, on-site field experiments were conducted to collect data from 20 construction workers. The performance of the developed model was assessed with leave-one-subject-out cross-validation (LOSOCV) to accommodate a wide range of new workers and their working conditions. According to the results, the work-at-height identification system is 96% accurate, while the safety hook attachment detection system is 86% accurate. The findings of this study fill knowledge gaps by providing ways of identifying workers working at height and detecting the fastening state of safety hooks in a non-invasive and objective manner. The results are expected to improve safety management at construction sites by minimizing the FFH risk for workers working at height.
This work was supported by the National Research Foundation of Korea (NFR) grant founded by the Korea Government (MSTI) (No. 2020R1G1A1004797 ). The authors wish to thank their industry partners for their help in data collection, as well as anonymous participants who participated in the data collection.
